Add support for latest AHRQ Elixhauser comorbidity calculation

AHRQ-Elixhauser/sas-parse/icd10cm_2020_1/get_lofmsdrg.R

@@ -0,0 +1,73 @@
+library(stringr)
+library(dplyr)
+
+# Make lofmsdrg from comformat_icd10cm_2020_1.txt
+download.file(
+  url = "https://www.hcup-us.AHRQ.gov/toolssoftware/comorbidityicd10/comformat_icd10cm_2020_1.txt",
+  destfile = "AHRQ-Elixhauser/sas-parse/icd10cm_2020_1/comformat_icd10cm_2020_1.txt"
+)
+
+# Read in data
+sas_path <- "AHRQ-Elixhauser/sas-parse/icd10cm_2020_1/comformat_icd10cm_2020_1.txt"
+sas_AHRQ_raw <- readLines(sas_path)
+
+
+# Helper functions to format msdrgs
+convert_interval = function(interval) {
+  split_interval = str_split(interval, '-')[[1]]
+  if (length(split_interval)>1) {
+    c(as.numeric(split_interval[1]):as.numeric(split_interval[2]))
+  } else {
+    as.numeric(split_interval[1])
+  }
+}
+
+format_msdrg = function(x) {
+  x %>%
+    str_split(',') %>% # Separate intervals
+    unlist() %>% # Unlist 
+    str_trim() %>% # Trim whitespace
+    .[.!=""] %>% # Remove blanks
+    sapply(convert_interval) %>% # convert to numeric intervals
+    unlist() %>% # clean up
+    unname() %>% # clean up 
+    as.vector() # keep consistent with vectors
+}
+
+make_lofmsdrg <- function(sas_AHRQ_raw){
+  raw_msdrg = sas_AHRQ_raw[-(1:grep("ICD-10 MS-DRG V37 Formats", 
+                                    sas_AHRQ_raw))] # Skip to MS-DRG
+  raw_msdrg = raw_msdrg[raw_msdrg!="" & 
+                          raw_msdrg!="Run;" & 
+                          raw_msdrg!="   "] # Drop empty and run
+  raw_msdrg = str_trim(raw_msdrg)
+
+  msdrg_labels = list()
+  msdrg_num_unformated = c()
+  for (i in raw_msdrg){
+    # Get value labels
+    if (grepl("VALUE", i)){
+      split_label = str_split(i, 'VALUE')[[1]][[2]] %>%
+        str_trim() %>%
+        str_split(" ")
+      msdrg_labels[[split_label[[1]][1]]] = list()
+      last_value = split_label[[1]][1]
+    }
+    if (grepl("\\d", i[1])){ # Extract numbers
+      msdrg_num_unformated = append(msdrg_num_unformated,
+                                    str_split(i, ' = ')[[1]][1])
+    }
+    if (grepl(';', i)){ # assign formatted MS-DRGs to label
+      msdrg_labels[[split_label[[1]][1]]] = format_msdrg(msdrg_num_unformated)
+      msdrg_num_unformated = c() # clear MS-DRG list for next loop
+    }
+  }
+  msdrg_labels
+}
+lofmsdrg = make_lofmsdrg(sas_AHRQ_raw)
+
+# Save lofmsdrg as .Rds
+saveRDS(lofmsdrg, 'AHRQ-Elixhauser/sas-formats/icd10cm_2020_1/icd10cm_2020_1_lofmsdrg.Rds')
+
+# Remove comformat_icd10cm_2020_1.txt
+file.remove(sas_path)

AHRQ-Elixhauser/sas-parse/icd10cm_2020_1/get_lofregex.R

@@ -0,0 +1,66 @@
+library(stringr)
+library(dplyr)
+
+download.file(
+  url = "https://www.hcup-us.AHRQ.gov/toolssoftware/comorbidityicd10/comformat_icd10cm_2020_1.txt",
+  destfile = "AHRQ-Elixhauser/sas-parse/icd10cm_2020_1/comformat_icd10cm_2020_1.txt"
+)
+
+
+# Example pattern to extract:
+# "D500",
+# "O9081",
+# "O99011",
+# "O99012",
+# "O99013",
+# "O99019",
+# "O9902",
+# "O9903"="BLDLOSS"        /*Blood loss anemia*/
+
+# Read in data
+sas_path <- "AHRQ-Elixhauser/sas-parse/icd10cm_2020_1/comformat_icd10cm_2020_1.txt"
+sas_AHRQ_raw <- readLines(sas_path)
+
+# Make list of lists for AHRQ codes to compare
+make_sas_list = function(sas_AHRQ_raw){
+  # Assigns ICD-10 codes to comorbidty labels from sas file located here:
+  # https://www.hcup-us.AHRQ.gov/toolssoftware/comorbidityicd10/comformat_icd10cm_2020_1.txt
+  # Omits /**** ICD-10 MS-DRG V37 Formats ****/
+
+  # Clean up readlines
+  sas_AHRQ_prep <- sas_AHRQ_raw[sas_AHRQ_raw!=""] %>% # Remove empty lines
+    .[-(1:18)] %>% # First 18 elements are extraneous
+    lapply(function(x) str_split(x,"\\/\\*")[[1]][1] ) %>% # Drop sas comments
+    unlist() %>%
+    str_trim() %>% # Trim white space
+    str_replace_all('\\"', "") %>% # Remove extraneous characters
+    str_replace_all(',', "") # Remove extraneous characters
+
+  AHRQ_list = list() # create empty list
+  temp_list = c() # placeholder for codes
+  for(l in sas_AHRQ_prep){
+    if(grepl("=", l, fixed=T)){
+      split_l = str_split(l,'=')[[1]]
+      temp_list = append(temp_list, split_l[1])
+      # AHRQ_list[[split_l[2]]] = str_c(temp_list, collapse="|")
+      # Must have ^ so that regex doesn't search for within-code substrings
+      AHRQ_list[[split_l[2]]] = paste0("^", str_c(temp_list, collapse="|^"))
+      temp_list = c()
+    } else {
+      temp_list = append(temp_list, l)
+    }
+    # Omit everything after wghtloss
+    if(l == "R636=WGHTLOSS"){
+      break
+    }
+  }
+  AHRQ_list # return the list
+}
+icd10cm_2020_1_lofregex = make_sas_list(sas_AHRQ_raw)
+
+# Save AHRQ_list object as RDS
+saveRDS(icd10cm_2020_1_lofregex, 
+        'AHRQ-Elixhauser/sas-formats/icd10cm_2020_1/icd10cm_2020_1_lofregex.Rds')
+
+# Remove comformat_icd10cm_2020_1.txt
+file.remove(sas_path)

AHRQ-Elixhauser/sas-parse/icd10cm_2021_1/get_mappings.R

@@ -0,0 +1,243 @@
+
+# Download file
+download.file(
+  url = "https://www.hcup-us.ahrq.gov/toolssoftware/comorbidityicd10/ElixhauserComorbidity_v2021-1.zip",
+  destfile = "AHRQ-Elixhauser/sas-parse/icd10cm_2021_1/ElixhauserComorbidity_v2021-1.zip"
+)
+# Unzip
+unzip("AHRQ-Elixhauser/sas-parse/icd10cm_2021_1/ElixhauserComorbidity_v2021-1.zip",
+      exdir = 'AHRQ-Elixhauser/sas-parse/icd10cm_2021_1/ElixhauserComorbidity_v2021-1')
+
+# Get raw SAS code line-by-line
+raw_format = readLines(
+  "AHRQ-Elixhauser/sas-parse/icd10cm_2021_1/ElixhauserComorbidity_v2021-1/Comorb_ICD10CM_Format_v2021-1.sas"
+  )
+
+# Remove quotes, commas, and whitespace
+trim_format = trimws(gsub(',', '', gsub('"', "", raw_format))) 
+# Remove 'proc' lines
+trim_format = trim_format[!grepl('Proc', trim_format)]
+# Remove 'run' lines
+trim_format = trim_format[!grepl(';', trim_format)]
+# Remove 'other' lines
+trim_format = trim_format[!grepl('other', trim_format)]
+
+# Separate vector by blank line
+format_list = split(trim_format[trim_format!=''], 
+      cumsum(trim_format=="")[trim_format!=''])
+
+# Remove extraneous
+format_list = format_list[3:length(format_list)] # Header stuff
+
+# Split into value groups
+new_values = unlist(
+  lapply(format_list, 
+       function(x) {
+         any(grepl('Value \\$', x))
+       }) 
+  )
+
+ElixhauserAHRQ2021Map = sapply(1:max(cumsum(new_values)),
+       function(x){
+         format_list[cumsum(new_values)==x]
+       })
+
+# Name the value groups
+names(ElixhauserAHRQ2021Map) = unlist(
+  lapply(ElixhauserAHRQ2021Map,
+       function(x){
+         strsplit(x[[1]][1], '\\$')[[1]][2]
+       })
+)
+
+# Drop 'value' elements in comfmt
+ElixhauserAHRQ2021Map$comfmt = sapply(ElixhauserAHRQ2021Map$comfmt, 
+       function(x) x[!grepl('Value', x)])
+
+# Get icd group names for comfmt
+names(ElixhauserAHRQ2021Map$comfmt) = sapply(ElixhauserAHRQ2021Map$comfmt, 
+       function(x){
+         strsplit(x[grepl(' = ', x)], ' = ')[[1]][2]
+       })
+
+# Drop ' = XXXX' from icd groups
+ElixhauserAHRQ2021Map$comfmt = sapply(ElixhauserAHRQ2021Map$comfmt, 
+       function(x){
+         x = sapply(x, function(x) {
+           strsplit(x, ' = ')[[1]][1]
+         })
+         names(x) = NULL
+         x
+       })
+
+# Drop " = 1" from poaxmpt and 'value' elements
+poaxmpt_names = names(ElixhauserAHRQ2021Map)[grepl('poa', names(ElixhauserAHRQ2021Map))]
+for (i in poaxmpt_names){
+  # Drop " = 1"
+  ElixhauserAHRQ2021Map[[i]] = unlist(
+    lapply(
+      strsplit(ElixhauserAHRQ2021Map[[i]][[1]],
+               ' = '),
+      function(x) {
+        x[[1]][1]
+      }
+    ) 
+  )
+
+  # Drop 'Value' elements
+  ElixhauserAHRQ2021Map[[i]] = ElixhauserAHRQ2021Map[[i]][!grepl(
+    'Value', ElixhauserAHRQ2021Map[[i]])]
+}
+
+# Define and save final comorbidities in AHRQ format:
+ElixhauserAHRQ2021Abbr = c(
+  'AIDS',
+  'ALCOHOL',
+  'ANEMDEF',
+  'ARTH',
+  'BLDLOSS',
+  'CANCER_LYMPH',
+  'CANCER_LEUK',
+  'CANCER_METS',
+  'CANCER_NSITU',
+  'CANCER_SOLID',
+  'CBVD',
+  'CHF',
+  'COAG',
+  'DEMENTIA',
+  'DEPRESS',
+  'DIAB_UNCX',
+  'DIAB_CX',
+  'DRUG_ABUSE',
+  'HTN_CX',
+  'HTN_UNCX',
+  'LIVER_MLD',
+  'LIVER_SEV',
+  'LUNG_CHRONIC',
+  'NEURO_MOVT',
+  'NEURO_OTH',
+  'NEURO_SEIZ',
+  'OBESE',
+  'PARALYSIS',
+  'PERIVASC',
+  'PSYCHOSES',
+  'PULMCIRC',
+  'RENLFL_MOD',
+  'RENLFL_SEV',
+  'THYROID_HYPO',
+  'THYROID_OTH',
+  'ULCER_PEPTIC',
+  'VALVE',
+  'WGHTLOSS'
+)
+
+# Define and save value labels (see Comorb_ICD10CM_Format_v2021-1.sas)
+ElixhauserAHRQ2021Labels = c(
+  'AIDS' = 'Acquired immune deficiency syndrome',
+  'ALCOHOL' = 'Alcohol abuse',
+  'ANEMDEF' = 'Deficiency anemias',
+  'ARTH' = 'Arthropathies',
+  'BLDLOSS' = 'Chronic blood loss anemia',
+  'CANCER_LEUK' = 'Leukemia',
+  'CANCER_LYMPH' = 'Lymphoma',
+  'CANCER_METS' = 'Metastatic cancer',
+  'CANCER_NSITU' = 'Solid tumor without metastasis, in situ',
+  'CANCER_SOLID' = 'Solid tumor without metastasis, malignant',
+  'CBVD' = 'Cerebrovascular disease',
+  'CBVD_NPOA' = 'Cerebrovascular disease, not on admission',
+  'CBVD_POA' = 'Cerebrovascular disease, on admission',
+  'CBVD_SQLA' = 'Cerebrovascular disease, sequela',
+  'CHF' = 'Congestive heart failure',
+  'COAG' = 'Coagulopthy',
+  'DEMENTIA' = 'Dementia',
+  'DEPRESS' = 'Depression',
+  'DIAB_CX' = 'Diabetes with chronic complications',
+  'DIAB_UNCX' = 'Diabetes without chronic complications',
+  'DRUG_ABUSE' = 'Drug abuse',
+  'HTN_CX' = 'Hypertension, complicated',
+  'HTN_UNCX' = 'Hypertension, uncomplicated',
+  'LIVER_MLD' = 'Liver disease, mild',
+  'LIVER_SEV' = 'Liver disease, moderate to severe',
+  'LUNG_CHRONIC' = 'Chronic pulmonary disease',
+  'NEURO_MOVT' = 'Neurological disorders affecting movement',
+  'NEURO_OTH' = 'Other neurological disorders',
+  'NEURO_SEIZ' = 'Seizures and epilepsy',
+  'OBESE' = 'Obesity',
+  'PARALYSIS' = 'Paralysis',
+  'PERIVASC' = 'Peripheral vascular disease',
+  'PSYCHOSES' = 'Psychoses',
+  'PULMCIRC' = 'Pulmonary circulation disease',
+  'RENLFL_MOD' = 'Renal failure, moderate',
+  'RENLFL_SEV' = 'Renal failure, severe',
+  'THYROID_HYPO' = 'Hypothyroidism',
+  'THYROID_OTH' = 'Other thyroid disorders',
+  'ULCER_PEPTIC' = 'Peptic ulcer disease x bleeding',
+  'VALVE' = 'Valvular disease',
+  'WGHTLOSS' = 'Weight loss'
+)
+
+ElixhauserAHRQ2021PreExclusion = c(
+  "AIDS",
+  "ALCOHOL",
+  "ANEMDEF",
+  "ARTH",
+  "BLDLOSS",
+  "CANCER_LYMPH",
+  "CANCER_LEUK",
+  "CANCER_METS",
+  "CANCER_NSITU",
+  "CANCER_SOLID",
+  "CBVD_SQLA",
+  "CBVD_POA",
+  "CBVD_NPOA",
+  "CBVD",
+  "CHF",
+  "COAG",
+  "DEMENTIA",
+  "DEPRESS",
+  "DIAB_UNCX",
+  "DIAB_CX",
+  "DRUG_ABUSE",
+  "HTN_CX",
+  "HTN_UNCX",
+  "LIVER_MLD",
+  "LIVER_SEV",
+  "LUNG_CHRONIC",
+  "NEURO_MOVT",
+  "NEURO_OTH",
+  "NEURO_SEIZ",
+  "OBESE",
+  "PARALYSIS",
+  "PERIVASC",
+  "PSYCHOSES",
+  "PULMCIRC",
+  "RENLFL_MOD",
+  "RENLFL_SEV",
+  "THYROID_HYPO",
+  "THYROID_OTH",
+  "ULCER_PEPTIC",
+  "VALVE",
+  "WGHTLOSS"
+)
+
+# Save list of format objects
+Elixhauser2021Formats = list(
+  ElixhauserAHRQ2021Map = ElixhauserAHRQ2021Map,
+  ElixhauserAHRQ2021Abbr = ElixhauserAHRQ2021Abbr,
+  ElixhauserAHRQ2021Labels = ElixhauserAHRQ2021Labels,
+  ElixhauserAHRQ2021PreExclusion = ElixhauserAHRQ2021PreExclusion
+)
+
+saveRDS(Elixhauser2021Formats, 
+        'AHRQ-Elixhauser/sas-formats/icd10cm_2021_1/Elixhauser2021Formats.Rds')
+
+# Remove .zip file
+file.remove(
+  "AHRQ-Elixhauser/sas-parse/icd10cm_2021_1/ElixhauserComorbidity_v2021-1.zip"
+)
+
+# Remove unzipped folder
+unlink(
+  'AHRQ-Elixhauser/sas-parse/icd10cm_2021_1/ElixhauserComorbidity_v2021-1',
+  recursive = T
+)